1. Introduction
Associated with knee symptoms and dysfunction, focal cartilage lesions are common in the general population. Hjelle et al. reported (osteo)chondral lesions (of any type) in 61% of their patients undergoing knee arthroscopy [
1]. These findings were confirmed by other studies, too [
2,
3,
4]. The tissue’s limited intrinsic healing capacity and the progressive nature of cartilage lesions warrant additional diagnostic and therapeutic efforts to prevent osteoarthritis (OA) and its great socioeconomic and personal disease burden.
While the aetiology of focal cartilage lesions is multifactorial, focal cartilage lesions are often the result of trauma [
5,
6]: Meniscus and anterior cruciate ligament (ACL) injuries bring about instability and predispose the joint to cartilage lesions [
5]. Similar dispositions are incurred by patellar dislocations. The prevalence rates of cartilage lesions in the patellofemoral joint are 71%, 82%, and 97% in acute, recurrent, and chronic dislocators, respectively [
6]. Other aetiologic factors are fractures, soft-tissue injuries, and repetitive microtraumatizations that result in surface incongruity, altered joint kinematics, and chronic degenerative changes, thereby predisposing to cartilage lesions, too. Consequently, posttraumatic OA (PTOA) accounts for nearly 12% of all cases of symptomatic OA in the United States [
7].
Due to its high soft tissue contrast and spatial resolution, non-invasiveness, and lack of radiation, Magnetic Resonance Imaging (MRI) is clearly the most powerful diagnostic tool of contemporary clinical medicine and the superordinate standard imaging modality for suspected joint and cartilage disorders [
8,
9,
10]. However, numerous studies have indicated the limitations of clinical-standard morphologic MRI techniques in the detection of cartilage lesions with variable sensitivities of 45% to 74% [
11,
12]. With the positive predictive value equally variable, morphologic MRI techniques are (i) not able to reliably indicate the presence (or absence) of cartilage lesions and (ii) particularly limited in detecting early, potentially reversible cartilage lesions. Consequently, quantitative MRI techniques such as T2 and T1ρ mapping have received ever-increasing scientific and clinical attention over the last decades [
13,
14]. These techniques quantify biophysical tissue properties on the compositional and ultrastructural level beyond mere morphology. Widely available on clinical MRI scanners and conveniently acquired with an additional scan time of 5 min, the addition of a T2 mapping sequence to a routine imaging protocol improved sensitivity in the detection of (early) cartilage lesions [
15]. T2 mapping is a robust, clinically and scientifically well-validated, and commonly used technique to assess cartilage status [
13,
14]. Moreover, T2 mapping is closely associated with relevant structural and compositional tissue features such as collagen content, collagen network organization and integrity, and water content [
16]. Consequently, a solid body of evidence has been collected that indicates the potential of T2 mapping in evaluating posttraumatic cartilage changes [
17,
18,
19,
20,
21,
22,
23].
Instead of merely quantifying T2 values of the superficial and deep cartilage zones in a pixel-wise manner, recent approaches have relied on more comprehensive post hoc approaches for image analysis such as texture feature analyses. During cartilage degeneration, collagen network integrity and proteoglycan content are lost [
24]. The increasing degrees of tissue disruption and disorganization translate to altered spatial distributions of T2 and may be quantified as markers of heterogeneity based on textural features. In degenerated cartilage, T2 values tend to be elevated with greater local heterogeneity [
17,
25] as has been demonstrated for cartilage lesions [
26], symptomatic OA and patients at risk of developing OA [
25,
27,
28], and after ACL injury [
29].
Despite this wealth of clinical knowledge, a basic understanding of the posttraumatic degenerative changes in cartilage and their imaging correlates is lacking. The present study’s objective was to contribute to this understanding by bringing together intact human articular cartilage, standardized injurious impaction loading with variable impaction energies, and T2 mapping and post hoc texture feature analysis. To this end, (histologically referenced) intact cartilage tissue was subject to impaction loading using a drop-tower device as an established model for inducing posttraumatic degenerative changes [
30,
31,
32], and imaged longitudinally to study these changes as a function of time and impaction energy, i.e., trauma severity. Our hypotheses were that (i) variable impaction energies induce variable progressive posttraumatic degenerative changes in cartilage and that (ii) these changes are reflected by the T2 maps and associated descriptive statistics and texture features.
4. Discussion
The most important finding of this study is that advanced MRI acquisition and postprocessing techniques, i.e., quantitative T2 mapping and texture feature analysis, may be used to (i) differentiate the severity of supraphysiological impact injuries of cartilage and (ii) monitor post-traumatic degenerative changes.
Prior to its initiation, this study had been motivated by the
lack of basic research available on the association of T2 mapping and traumatic cartilage injury. Even though a solid body of clinical evidence is available to support the potential of T2 mapping in evaluating posttraumatic cartilage changes [
17], the changes in T2 related to traumatic injury are variable and inconsistent. In young adults with recurrent patellar dislocations during childhood, significant decreases in T2 were found in the superficial patellar cartilage zone of injured as compared to non-injured joints, which may be an early sign of cartilage pathology [
18]. After ACL reconstruction, inconsistent T2 changes relative to uninjured controls were found. At the one-year follow-up, T2 was not significantly elevated in one study [
19], while others found significant T2 elevations at the medial femoral cartilage after two [
20] and three years [
21]. In contrast, increases in T2 were associated with morphologic cartilage lesions [
22] as well as morphologically intact cartilage that is going to develop morphologic cartilage lesions in the years to come [
23]. This, of course, indicates that compositional changes -as assessed by T2 mapping- precede the development of morphologic cartilage lesions and underscores the potential of T2 mapping to identify cartilage regions at risk of incipient degeneration.
This study clearly demonstrates that
injurious cartilage impaction is associated with increasing T2 values as a function of impaction energy level and time. Based on the sensitivity profile of T2 versus structural and compositional cartilage properties, these increases reflect numerous single-impact-associated posttraumatic changes in cartilage [
32,
41,
42,
48,
49,
50,
51]. The literature data indicate that these changes include surface damage, loss of proteoglycans and collagen network integrity, as well as chondrocyte death. These changes closely resemble degenerative changes in OA, are therefore often referred to as traumatic OA-like changes [
42], and provide the degenerative correlates of altered T2 values. For the sake of comparability, impaction energy levels were chosen in line with earlier studies and, mechanistically, the induction of cartilage damage by dropping weights from defined heights has been thoroughly validated before [
32,
41,
48]. Nonetheless, in this study, histologic reference indicated impaction-energy-associated surface disintegration and incipient-to-moderate proteoglycan depletion, thereby confirming the mode of action within the framework of this study.
On the tissue level, the impaction-induced
increases in T2 may be secondary to numerous posttraumatic changes that are excellently reviewed in [
52]. For once, proteoglycans (and -in parts- collagen) are lost secondary to the collagen network damage. Lower proteoglycan and collagen contents are associated with higher T2 values [
53]. For another, collagen network disintegration is induced directly by mechanical disruption and indirectly by subsequent enzymatic degradation. These processes contribute to increased tissue water content and tissue swelling, which are associated with higher T2 values, too [
33], as well as increased collagen fibre disorientation and anisotropy. Secondary to impaction, the percentage of fibres oriented at magic angle, i.e., at 55° to the main magnetic field, may be elevated, thereby increasing T2 values, too [
54]. Yet, even though these mechanisms are plausible, it remains unclear which exact compositional or (ultra)structural mechanism is primarily behind the prominent increases in T2.
Beyond mean T2 values, this study focused on
radiomic texture features, too, as refined imaging biomarkers of cartilage trauma. Again, changes were clear and significant after HIMP exposure, while they were moderate and only tended towards significance after LIMP exposure. Across the spectrum of texture features assessed, variance and metrics of contrast and orderliness were significantly increased (i.e., contrast, variance) and decreased (i.e., homogeneity, energy) as a function of impaction energy. Additionally, these features were different between the various time points, indicating lower textural uniformity and growing structural disorder. These changes may be considered a sign of cartilage damage, too, as described before [
25,
26,
44].
Notably, changes in T2, i.e., absolute values and texture features, were subject to gradual and drastic
changes over time. These aspects may be explained by the concurrence of posttraumatic changes in cartilage that are induced either immediately through the impaction itself or delayed through cell death and the induction of matrix-degrading enzymes. These processes reduce biosynthetic capacity or bring about progressive degradation that may explain the gradual alterations in T2 characteristics. Most likely, these processes are at the root of the hyperintense bands that traversed the cartilage sample in the transitional zone parallel to the subchondral lamella. Observed after both low- and high-energy exposure, similar histologic changes have been reported before [
32,
41]. Following impaction of bovine cartilage, Jeffrey et al. found horizontal fissures in the transitional zone that they hypothesized to be due to deflection of the extracellular matrix with partial delamination of the upper and lower tissue portions. Other studies also demonstrated sub-surface intra-tissue damage and chondrocyte death prior to surface disintegration [
55]. The band’s progression in terms of size and signal characteristics was clearly associated with impaction energy level and may thus reflect the ongoing structural and compositional changes of traumatized cartilage.
Notably,
zonal changes of the superficial and deep tissue layers were roughly similar in terms of relative changes in T2. Considering the direct impaction of the cartilage surface, intuitively, one would expect larger changes of superficial than deeper zones. Yet, the viscoelastic nature of cartilage, its unique compressive properties and tight attachment to the underlying subchondral bone provide efficient mechanisms for absorption of physiologic and supraphysiologic loads throughout the entire tissue depth [
56]. Once the subchondral bone is removed, these mechanisms of load distribution and dissipation are disrupted and the protective effect is lost [
41].
This study has numerous
limitations. First, the experimental in vitro design necessitated excision and preparation of cartilage and its prolonged incubation, thereby limiting the clinical translatability of our findings. Prolonged incubation in media may artificially increase tissue hydration, thereby increasing T2 values. As this was observed for otherwise unaffected control samples, these gradual increases in T2 provide the background against which the impaction-induced changes must be considered. Additionally, scanning was performed at room temperature, again affecting T2 values. As the MRI measurements were performed in a standardized manner, this bias may be considered systematic. Additional studies using in situ human whole-knee joint configurations and more physiologic impaction methods are thus required to confirm our findings. As an additional caveat, however, resultant intra-tissue changes secondary to impaction are largely dependent on the experimental framework conditions. Standardized impaction of cartilage samples induces more severe changes than similar impaction of intact joints [
57]. Second, human cartilage samples were obtained from knee joints undergoing joint replacement. Despite our best efforts to ascertain tissue quality by macroscopic evaluation and baseline histology, the pre-existent degeneration of the tissue brought about by chronic mechanical and inflammatory disease processes is clearly uncontrolled and may have altered the tissue’s susceptibility to impaction loading. Even though the longitudinal study design allows for consistent intra-sample referencing and reduces this type of bias, tissue variability clearly affects tissue susceptibility and outcomes of impaction. Future studies should therefore use ‘truly’ healthy cartilage through alternative tissue sources such as amputations or organ-donor networks to realize improved tissue quality control. Third, despite selecting cartilage from the lateral femoral condyles for reasons of topoanatomic consistency, human cartilage thickness is largely different between individuals [
58] and may affect load distribution and dissipation in the tissue. Fourth, the cartilage samples surfaces were oriented parallel to the main magnetic field B0 and, consequently, the majority of collagen fibres of the deep and transitional zones were oriented at 90° to B0, thereby affecting T2 quantification. T2 values may increase due to the magic angle effect as collagen fibre orientation changes relative to B
0 after injurious impaction, particularly in the deep tissue layer [
54] Although measurement conditions were standardized (using a dedicated device) and inter-sample variability in position and configuration thus decreased, this aspect needs to be considered, too, prior to any clinical translation. Fifth, this study only focused on T2 mapping, while other imaging markers, e.g., T1ρ, T1, T2 *, sodium, gagCEST (glycosaminoglycan Chemical Exchange Saturation Transfer), or contrast-enhanced techniques such as dGEMRIC (delayed gadolinium-enhanced MRI of cartilage) are of potential value in assessing cartilage structure and composition [
14,
34,
38,
59,
60,
61,
62,
63,
64]. Of these, T1ρ mapping is of additional and complementary value to T2 mapping because of distinct biophysical properties [
65]. Yet, the exact sensitivity profile of T1ρ remains to be determined with proteoglycan, collagen, and water content as well as collagen fibre orientation potentially contributing to T1ρ relaxation characteristics. As of today, T1ρ seems to indicate the cartilage tissue’ macromolecular configuration [
53,
66,
67,
68,
69]. In posttraumatic contexts, the sensitivity of T1ρ to changes in the tissue’s solid and fluid constituents and its mechanical condition [
38,
70,
71,
72,
73] may be of value in future pre-clinical and clinical studies. Sixth, this study did not include dedicated compositional reference measures and, consequently, no advanced quantification of proteoglycan or collagen content that would have allowed spatially resolved associations of T2 maps and compositional measures. Prior to any clinical translation, these associations ought to be clarified in posttraumatic contexts and beyond.